Abstract
Objective: To evaluate the predictive validity and potential clinical relevance of the DriveSafe DriveAware assessment for predicting on-road driving outcomes in medically at-risk drivers. Design: Prospective cohort study. Setting: Fitness-to-drive assessment pathway involving a driver licensing authority, medical review clinic, and occupational therapy driver assessors in the Australian Capital Territory, Australia. Participants: A convenience sample of 138 drivers referred for occupational therapy driving assessment because of medical fitness-to-drive concerns. Methods: Participants completed the DriveSafe DriveAware, comprising DriveSafe and DriveAware subtests, before an on-road driving assessment. On-road outcomes were dichotomised as pass (unrestricted licence) or fail (conditional licence, reassessment, or licence cancellation). Receiver operating characteristic analyses, Youden's Index, and classification accuracy metrics were calculated. Results: DriveSafe DriveAware demonstrated strong classification performance, with 86% sensitivity, 100% specificity, 100% positive predictive value, and 93% overall accuracy. Eighty-seven percent of participants were classified as Likely to Pass or Likely to Fail , with 13% classified as Further Testing . No participants classified as Further Testing achieved an unconditional pass. DriveSafe and DriveAware subtests showed excellent discrimination (area under the curve = 0.96 and 0.95, respectively). Optimal cut-offs were broadly consistent with previously proposed thresholds. Drivers who failed the on-road assessment were older and had lower DriveSafe and DriveAware subtest scores than those who passed. Conclusion: DriveSafe DriveAware showed strong predictive validity in this sample and may support triage and clinical decision-making within fitness-to-drive assessment pathways.
Funding Information
  • Sam Ryan Memorial Award 2017 and the Cure Brain Cancer Foundation (Sam Ryan Memorial Award 2017)
  • Canberra Hospital Private Practice Fund

This publication has 28 references indexed in Scilit: